| name | memory-test |
| description | Run E2E memory pipeline test โ embed text, store in Qdrant, search and retrieve. Use to validate the memory system is working. |
| allowed-tools | Bash |
Memory Pipeline E2E Test
Run the full embed โ store โ search pipeline:
-
Start port-forward (if not already running):
kubectl port-forward svc/qdrant -n vector 6333:6333 &
sleep 2
-
Generate embedding:
curl -s http://10.10.10.10:30001/v1/embeddings \
-H "Content-Type: application/json" \
-d '{"model":"default","input":"Kaizen memory pipeline test at '"$(date -Iseconds)"'"}'
Verify: 4096 dimensions returned.
-
Store in Qdrant:
Use the embedding from step 2 to upsert a point into the episodic collection with a random UUID and timestamp payload.
-
Search:
Generate a new embedding for a related query and search the episodic collection. Verify score > 0.5.
-
Report:
| Step | Status | Details |
|---|
| Embedding | ? | dims, latency |
| Store | ? | collection, point ID |
| Search | ? | score, matches |
-
Cleanup port-forward:
kill %1 2>/dev/null